JPEG XR – Standardizing Microsoft HD Photo

A bit of good standards news today. Earlier today there was a press release from the Joint Photographic Expert Group (JPEG) and one from Microsoft both announcing that the specification for Microsoft’s HD Photo as been submitted to JPEG. The result, hopefully, will be the next generation of JPEG called JPEG XR. We are working collaboratively with the JPEG community and hope to see this result in the next wave of digital image technologies.

The benefits of this potentially extend to all digital camera users, to camera manufacturers, printer manufacturers, and to anyone building apps that manipulate digital images. (Think mapping, imaging, etc.)

JPEG XR will part of a larger piece of work called the JPEG Systems which the JPEG organization is doing to standardize systems integration technologies for digital imaging technologies.

A few news stories on this today:

CNET – Shankland

eWEEK – Galli

CNN Money

In the blogosphere, a few good places to watch this space:

Bill Crow – has a blog dedicated to HD Photo

Microsoft PhotoBlog – cool site to watch if you are into digital photography


If you are curious about HD Photo from the tech perspective, I found this link. Also, here is some basic information about HD Photo submission from Microsoft:

Q. Why did Microsoft decide to submit HD Photo for consideration as a standard?

A:  More efficient memory utilization, better quality pictures and prints, more flexible editing, and support for emerging image rendering and interaction innovations are all features that are highly valued by users.  Microsoft believes that HD Photo will address these and other current and emerging needs of consumer and professional digital photographers. We decided to submit HD Photo for consideration as a standard because standardizing HD Photo will expand access to this cutting edge technology and foster interoperability with other related standards and innovations.

Q: What are the key benefits of the HD Photo file format?

A. This file format introduces support for High Dynamic Range (HDR), a major and fundamental new development in digital imaging.  The key benefits include:  1) preserving a far greater range of original image content, thus enabling the highest quality exposure and color adjustment; 2) delivering  better efficiency with fewer damaging artifacts and scalable to lossless; and 3) the ability to decode only the information needed for any resolution or region and to manipulate the image as compressed data. 

Comments (19)

  1. Adrian Ford says:

    HD Photo (previously WM Photo or Windows Media Photo) is one of the image formats supported within the

  2. HD Photo (previously WM Photo or Windows Media Photo) is one of the image formats supported within the

  3. Andrew Sayers says:

    This post is sort of a follow on from my previous post in "Ecma Open XML and the Portuguese National Body" – my question is relevant here though, and I’d rather not pollute the useful conversation going on over there.

    Jason, I hope it helps to know you’re not the only one spending extra time on this – my inability to express the questions I’d like to ask has been eating away at me since my last post, and now I think I might have got it:

    Would you say that Microsoft sees the purpose of the ISO as more about documenting current practice, rather than trying to find theoretically perfect solutions to problems?  I realise this isn’t a black-and-white issue, and pragmatists will always do a little of both, but the only thing I can think of that explains Microsoft’s behaviour is if the company puts a far lower value on elegance than I’m used to seeing.

    To be clear, I don’t necessarily see that as a bad thing – for example, the W3C did some of its best work when it was just documenting current practice.  It’s just that it’s important to have a strong model of Microsoft’s corporate thought process in order to have a useful conversation about its behaviour.

    – Andrew

  4. Simon Phipps says:

    Hi Jason – I took a little look around but counldn’t find an answer to this quastion. Will this format be covered by a patent pledge of sufficient quality to give open source develoeprs confidence to implement it? RF alone is no use if there are other obstacles (not sublicenceable or essential-claims-only for example).

  5. jasonmatusow says:

    Hi Simon, I hope you are having a good summer.

    I will check on the the terms for this submission and will get back to this comment thread.

    I have a question for you about this though. There are many, many, many specs that are implemented in free software today that cleary conflict with the terms associated with the spec. Essentially, any RAND-based spec conflicts with the GPL (Unicode, Firewire, etc. etc etc.). I am guessing (please correct me if I’m wrong) that the existing JPEG spec is probably implemented in Free Software already. I ask this because I wonder at the direction things are going. Is it that because of the Free Software model – all specifications must be under some form of OSP (patent promise/pledge/CNS) from now on? I’m guessing that there are rights holders out there who would prefer to retain the right to choose what terms they want to convey their rights under. MS has used the OSP – but that does not mean that all specs will go under that model.

    Anyway – I will find out and let you know.



  6. Rick Jelliffe says:

    Many ISO specs reflect external technologies. For example, the ISO Linux ABI standard, ISO C#, ISO ODF, ISO PDF/X. In those kind of cases, the quality of the standard is mainly determined by how well it documents the existing technology, warts and all, rather than being anything to do with perfection.

    Semi-failures such as OSI, ODA and HyTime have demonstrated to many people that "blue-sky" standards are trouble and rarely successful, and so it is much more common to look at external technologies. Where there is no dominant technology in the market, then bringing together existing vendors is a good approach, and hybrids sometimes work, but usually even then normally you expect that one technology will be adopted as the base.

    Some ISO committees, such as ISO/IEC JTC1 SC34, have a policy to only adopt royalty-free (or preferably patent-free) technologies. Part of the process of bringing a standard to the table is clarifying the patent position of the draft. Personally I think it is a mistake to standardize technologies on a RAND basis: they should be free. Technologies such as MPEG have been hindered by this. Companies get stars in their eyes, because they see it as a way to get a licensing stream.

    Of course, under those JEDEC-related cases, a company that participates in a standard with hidden IP or claiming that they will license their IP free, cannot turn around after the standard is finalized and claim license fees (unless they pull out before the end.)

  7. Simon Phipps says:

    Thanks, Jason.

    You’re right, there are plenty of standards extant that have terms that cause Free software developers problems. While some developers may implement in spite or in ignorance of this, the emerging market is one which demands transparent and open terms. Sun is making patent covenants that give Free software developers confidence to implement[1] and I’d encourage Microsoft (and indeed IBM) to follow suit.


  8. jasonmatusow says:

    Rick –

    I think there is a good deal of misunderstanding out there (I am not meaning to insinuate anything about your understanding) about RAND. Reasonable and Non-Discirminatory terms actually represent a group of ideas designed to encourage participation in, and adoption of standards. Royalties are only 1 of many ideas in RAND. Also, it has been common practice for years and years to have standards that bear RAND royalty terms yet have no fees associated with them. The lawyers keep the royalty terms in there as a defensive mechanism. There are other situations where royalties are collected from standards (look at Qualcomm’s business model to see how this may be used). Over time, a class of RAND terms with zero royalties (some call RANDz) has cropped up. Under these term, it is clearly stipulated that there will be no royalties associated with the patents in the spec. BUT, the other RAND elements like field of use restrictions, defensive suspension, etc. are maintained. Now, the next step towards openness is the CNS/Patent Pledge from contributors. Even there, complexity is still possible in that the standards body may have certain required IP terms (most probably based on RAND) even though the contributor has provided even more liberal terms.

    Again, for me – patents can be both blockers and facilitators for collaboration. I want to see standards stemming from strong innovation. If you follow the logic that I commented on with Simon, that because free software development is out there ALL standards must be under patent pledges etc – that may become an active disincentive for innovators to contribute to standards bodies.

    Innovation dead zones concern me. There has to be a way to balance these issues and find mechanisms that respect multiple approaches.


  9. jasonmatusow says:

    Simon –

    You guys have done great thinking around this area for a long time. You and your team’s work on things like the CDDL, CNS, etc. show a long-term committment to a set of both ideals and approaches to software development within the busiens context. I have nothing but respect for that.

    But, that thinking is done within the context of your business model and the realities of your business. The effect of Linux on the value prop of Solaris must have had an effect on your thinking (maybe for the better – no judgement from me on that). If you have a pure software business (as do many ISVs, MS included), you will have a different view.

    Free software development is compelling in some ways, and problematic in others. It is not the model that all wish to follow and thus finding mechanisms to bridge multiple models should be found.

    You say "some developers may implement in spite or in ignorance" of the conflict between RAND terms and GPL – but I don’t buy that. It is a question of not addressing it on purpose. IBM, Intel, HP, Red Hat, Novell – these are firms that clearly have deep sophistication when it comes to software development. IP, standards, etc. etc. There are certain things that are purposely overlooked because they are inconvienent. For example – licensing conflicts in Linux. All reciprocal licenses are by defnition incompatible yet somehow Linux carries dozens of incompatible licenses and no one cares. The FSF who has very high expectations of others – allow this flagship FLOSS technology to go without challenging it becuase it suits their needs to have it be that way.

    Microsoft is pushing more and more specs out under the OSP ( and we will continue to think deeply about what types of specs will use this model. But it will not be all of our contributions. Every contribution is considered individually – they are business decisions and all factors get looked at.

    Thx again for the comment.


  10. jasonmatusow says:

    Simon –

    Just verified the terms for the submission of JPEG XR. RANDz.

    For those of you not familiar with this, it is worth noting that this is the most permissive option available under JTC-1 rules. Under these terms, if the spec becomes a JTC-1 standard, the intent is to promote widespread adoption on various platforms.

    I would recommend that anyone curious about this go and look at the IPR policies for JTC-1.



  11. Andrew Sayers says:


    Thanks for the pointers about perfection vs. reality – it helps to deepen my understanding of the issue, and the divide between people on it.

    In particular, it suggests to me the OOXML debate represents a classic example of a fight between people who prefer to reason from precedent and those who start from pure logic.  If that really is the case, then in order to have a meaningful discussion with idealists, you’ll need to be much more explicit about the validity and benefits of reasoning from precedent, and you’ll find this will be a recurring theme throughout any sufficiently large standardisation debate (which JPEG XR might become).

    For example, I’ve seen a lot of defencive counters to the "OOXML is just a binary format wrapped in angle brackets" argument, but I’ve yet to see the argument made that this is actually a good thing – that you’re adding XML to something that’s known to work, rather than creating something new that you suspect will work.

    – Andrew

  12. Simon Phipps says:

    Thanks for looking into that, Jason. Is that RANDZ license sublicensable without reference to Microsoft?  It needs to be if open source developers are to implement it – one man’s "Reasonable" is another man’s "Restrictive", so "RAND" can mean more than one thing…

  13. Simon Phipps says:

    Jason:  Not sure if you missed my earlier comment. Is it possible to review the IPR terms for the JPEG XR submission from Microsoft please? I’m interested to know if the rights are sublicensable without reference to Microsoft since a $0 fee is not sufficient to allow FOSS developers to implement freely.

  14. jasonmatusow says:

    Simon – I will send mail internally on this and get back to you, and this blog thread. I would point out to you that your statement is both correct and a red herring at the same time. Unicode (for example) I believe is implemented in Free Software, but the spec is under RAND terms. RAND has much more than just royalties in it (as you well know), and a few of those terms conflict with the GPL (and probably with a few other reciprocal licenses such as the Sun licenses). Yet, people go ahead and implement those things all the time. So – there is a rub here. The idea that someone chooses to use a Free Software license is not the responsibility of the standards body, nor of the contributor. It is the choice of the implementer. If there is a conflct there, it is incumbant on the implementer to find a way to both implement the spec (with the associated IPR terms) and keep true to other legal commitments they have. That said, you see a wide range of activities being undertaken by vendors to try to help bridge this gap (our own OSP for example).

    I will go ask – but this is not a simple question/answer.


  15. jasonmatusow says:

    Simon – turns out I did ask…and simply missed the email in the mailstrom that is my inbox.

    We have submitted this spec under the most permissive option available under the JTC-1 rules. The terms are RANDz – we are not placing any royalties on the patents associated with this spec. Our patent grants to JPEG XR will promote widespread adoption on multiple platforms. One of the things suggested to me by my legal folks is that a) I’m not a lawyer and should not be giving direct advice on the terms in my blog, and b) that this all falls under JTC-1 rules, and you should go to those rules to look into any concerns you may have.



  16. Dave S. says:


    Are the specific patents listed in the submission?

    Can’t a RAND licensor come back to say that offering JPEG-XR editing is worth, say $20 per sold application? Is that unreasonable? Is 5$ unreasonable – $2? How low is that threshold?  

    I think the Free software license choice is made by the contributor, not the implementor. An implementor cannot unilateraly decide that the basis for development (JPEG-XR) is now uninhibited from the requirements under any particular license. It already has to meet those requirements. If Microsoft, in this case, does not clarify their position relative to use, then it’s an effective wall to such development.

    I do notice that the "Microsoft Open Specification Promise" is a convenant that ensures no patent countersuit.

    It also looks as if it forces patent sharing on the part of participants to the extent that if an implementor of  a covered item creates and patents a method for dealing with said covered item that Microsoft is free to use that patent in their own competing product. The alternative is the implementor can countersue, at which time the covenant is broken and any MS patented material is no longer available to the implementor.

  17. jasonmatusow says:

    Dave – RAND stands for "Reasonable And Non-Discriminatory, RANDz" stands for "Reasonable And Non-Discriminatory with Zero royalties." RAND(z) has more than just royalty terms associated with it. This is a legal construct built on more than 100 years of industrial standardization (and litigation/precedent). So, in this case Microsoft made a RANDz contribution – this means that the terms clearly state zero royalties. We will not collect royalties on our IP in that spec. BUT – HUGE CAVEAT – JPEG is a big organization with many contributors, and it may be that a given spec carries royalties from some other contributor. Also, there is nothing to say that the final spec won’t touch on IP held by some third party who was not in the standardization process to begin with. So, each contributor has to decide what terms they are comfortable with, AND have to function within the rules of the standards body (this is why IPR policies are such important things).

    Don’t confuse contribution terms with implementation terms. The spec is a document – it is not source code. When an implementer the goes and builds code based on the spec  – that code is placed under a license of the implementer’s choice. If that license conflicts with the terms of the IP placed into the implementation – then it is the responsibility of the implementer to deal with that.

    This is exactly why we have the OSP and IBM their patent promise, and Sun their Covenant Not To Sue – these are all mechanisms to try and bridge the world of IP holders with the world of Free Software development. At the same time, those choosing to do Free Software development need to also work hard to bridge back to the framework of intellectual property rights. (This is why the Mozilla Public License was written for example, or the CPL, EPL, or any other GPL-like license as people attempted to find that bridge). Standards simply add another complex wrinkle to the discussion.

    Hope that helps.


  18. Dave S. says:

    Again, are the specific patents that apply to the RANDz listed in the Microsoft submission?

    And, doesn’t the agreement mean that any software that handles this format which might be construed as using a Microsoft patent, effectively give Microsoft free access to any of the using companies patents?